# Computations
import numpy as np
import pandas as pd
import pickle

# preprocessing
from sklearn import preprocessing
import re

# Visualisation libraries

## Text
from colorama import Fore, Back, Style
from IPython.display import Image, display, Markdown, Latex, clear_output

## progressbar
import progressbar

## plotly
from plotly.offline import init_notebook_mode, iplot 
import plotly.graph_objs as go
import plotly.offline as py
from plotly.subplots import make_subplots
import plotly.express as px

## seaborn
import seaborn as sns

## matplotlib
import matplotlib.pyplot as plt
from matplotlib.patches import Ellipse, Polygon
from matplotlib.font_manager import FontProperties
import matplotlib.colors as mcolors
plt.style.use('seaborn-whitegrid')
plt.rcParams['axes.labelsize'] = 14
plt.rcParams['xtick.labelsize'] = 12
plt.rcParams['ytick.labelsize'] = 12
plt.rcParams['text.color'] = 'k'
%matplotlib inline

import warnings
warnings.filterwarnings("ignore")

Home / Financial Analysis and Modeling

Statlog (German Credit Data) Dataset

• Preprocessing
• Exploratory Data Analysis
• Modeling: PyTorch Multi-layer Perceptron (MLP) for Binary Classification
• Modeling: Gradient Boosting Classifier

Statlog (German Credit Data) Dataset

In this article, we analyze the UCI Statlog (german credit data) from Kaggle.com.

Context

The original dataset contains 1000 entries with 20 categorial/symbolic attributes prepared by Prof. Hofmann. In this dataset, each entry represents a person who takes a credit by a bank. Each person is classified as good or bad credit risks according to the set of attributes. The link to the original dataset can be found below.

Content

It is almost impossible to understand the original dataset due to its complicated system of categories and symbols. Thus, I wrote a small Python script to convert it into a readable CSV file. Several columns are simply ignored, because in my opinion either they are not important or their descriptions are obscure. The selected attributes are:

• Age (numeric)
• Sex (text: male, female)
• Job (numeric: 0 - unskilled and non-resident, 1 - unskilled and resident, 2 - skilled, 3 - highly skilled)
• Housing (text: own, rent, or free)
• Saving accounts (text - little, moderate, quite rich, rich)
• Checking account (numeric, in DM - Deutsch Mark)
• Credit amount (numeric, in DM)
• Duration (numeric, in month)
• Purpose (text: car, furniture/equipment, radio/TV, domestic appliances, repairs, education, business, vacation/others)

Path = 'Statlog_Dataset/german_credit_data.csv'
Data = pd.read_csv(Path, index_col=0)
display(Data)
Target = 'Risk'
Labels = [x.title() for x in Data[Target].unique()]

def Data_Plot(Inp, Title, W = None):
    data_info = Inp.dtypes.astype(str).to_frame(name='Data Type')
    Temp = Inp.isnull().sum().to_frame(name = 'Number of NaN Values')
    data_info = data_info.join(Temp, how='outer')
    data_info ['Size'] = Inp.shape[0]
    data_info['Percentage'] = 100 - np.round(100*(data_info['Number of NaN Values']/Inp.shape[0]),2)
    data_info = data_info.reset_index(drop = False).rename(columns = {'index':'Features'})
    #
    fig = px.bar(data_info, x= 'Features', y= 'Percentage', color = 'Data Type',
                 text = 'Percentage',
                 color_discrete_sequence = ['PaleGreen', 'LightCyan', 'PeachPuff', 'Pink', 'Plum'],
                 hover_data = data_info.columns)
    fig.update_layout(plot_bgcolor= 'white', legend=dict(x=1.01, y=.5, traceorder="normal",
                                                         bordercolor="DarkGray", borderwidth=1))
    if not W == None:
        fig.update_layout(width = W)
    fig.update_traces(texttemplate= 10*' ' + '%%{text}', textposition='inside')
    fig.update_traces(marker_line_color= 'Black', marker_line_width=1., opacity=1)
    fig.update_layout(title={'text': '<b>' + Title + '<b>', 'x':0.5,
                                     'y':0.90, 'xanchor': 'center', 'yanchor': 'top'})
    fig.show()
    return data_info
    
def dtypes_group(Inp, Dict = False):
    Temp = Inp.dtypes.to_frame(name='Data Type').sort_values(by=['Data Type'])
    Out = pd.DataFrame(index =Temp['Data Type'].unique(), columns = ['Features','Count'])
    for c in Temp['Data Type'].unique():
        Out.loc[Out.index == c, 'Features'] = [Temp.loc[Temp['Data Type'] == c].index.tolist()]
        Out.loc[Out.index == c, 'Count'] = len(Temp.loc[Temp['Data Type'] == c].index.tolist())
    Out.index.name = 'Data Type'
    Out = Out.reset_index(drop = False)
    Out['Data Type'] = Out['Data Type'].astype(str)
    
    if Dict:
        Out = dict(zip(Out['Data Type'], Out['Features']))
    return Out

_ = Data_Plot(Data, Title = 'Statlog Dataset', W = 800)
dType = dtypes_group(Data, Dict = True)

	Age	Sex	Job	Housing	Saving accounts	Checking account	Credit amount	Duration	Purpose	Risk
0	67	male	2	own	NaN	little	1169	6	radio/TV	good
1	22	female	2	own	little	moderate	5951	48	radio/TV	bad
2	49	male	1	own	little	NaN	2096	12	education	good
3	45	male	2	free	little	little	7882	42	furniture/equipment	good
4	53	male	2	free	little	little	4870	24	car	bad
...	...	...	...	...	...	...	...	...	...	...
995	31	female	1	own	little	NaN	1736	12	furniture/equipment	good
996	40	male	3	own	little	little	3857	30	car	good
997	38	male	2	own	little	NaN	804	12	radio/TV	good
998	23	male	2	free	little	little	1845	45	radio/TV	bad
999	27	male	2	own	moderate	moderate	4576	45	car	good

1000 rows × 10 columns

Preprocessing

Dealing with NaN values.

## Categorical Data
# Filling Nan values with None
Data[dType['object']] = Data[dType['object']].fillna('None')

Data[dType['object']] = Data[dType['object']].applymap(lambda x: x.title())
for Feat in dType['object']:
    Data[Feat] = Data[Feat].str.replace('Tv','TV')
    
Data.columns = [x.title() for x in Data.columns]    
_ = Data_Plot(Data, Title = 'Statlog Dataset', W = 800)

Data.to_csv (Path.split(".")[0]+'_Mod.csv', index = None, header=True)

Modifications

# A copy of the dataset
df = Data.copy()
# Feat Dictionary
Feat_Dict = {}

def Distinct_Observations(Inp, Target = Target, Featured_Columns = None, YL = None):
    if Featured_Columns == None:
        Featured_Columns = list(set(Data.columns) - {Target})
    Temp = Inp[Featured_Columns].nunique()
    fig = go.Figure([go.Bar(x=Temp.index, y=Temp.values)])
    fig.update_traces(marker_line_color= 'Navy', marker_line_width=1, opacity=1, showlegend = False)
    fig.update_layout(legend_orientation='v', plot_bgcolor= 'white', height= 450, width= 980,
                     title={'text': '<b>' + 'Distinct Observations in Each Column' + '<b>', 'x':0.5,
                            'y': 0.92, 'xanchor': 'center', 'yanchor': 'top'},
                     yaxis_title='Frequency')
    fig.update_xaxes(showline=True, linewidth=1, linecolor='Lightgray', mirror=True,
                         zeroline=False, zerolinewidth=1, zerolinecolor='Black',
                         showgrid=False, gridwidth=1, gridcolor='Lightgray')
    fig.update_yaxes(showline=True, linewidth=1, linecolor='Lightgray', mirror=True,
                     zeroline=True, zerolinewidth=1, zerolinecolor='Black',
                     showgrid=True, gridwidth=1, gridcolor='Lightgray')
    if not YL == None:
        fig.update_yaxes(range =[0, YL])
    fig.show()
    return Temp
Table = Distinct_Observations(df, Featured_Columns = None, YL = 1000)

def FeatBins(Inp, Bins, replace = True, String = True):
    Bins = [int(x) for x in Bins]
    Out = pd.cut(Inp, bins = pd.IntervalIndex.from_tuples([(x, y) for x, y in zip(Bins[:-1],Bins[1:])]))
    Temp = np.sort(Out.astype('str').unique())
    Dict = dict(zip(Temp, np.arange(len(Temp))))
    if replace:
        Out = Out.astype('str').replace(Dict)
    if String:
        Out = Out.astype('str')
        try:
            Out = Out.str.replace(pat = '(-1', repl = '[0')
        except:
            pass
    return Out

def BinPlot(Feat, Bins, xLim, Inp = Data, W = 600, TableColors = ['Navy','White']):
    fig = make_subplots(rows=2, cols=1, vertical_spacing = 0.1, row_heights=[.1, .9],
                        specs=[[{"type": "xy"}], [{"type": "table"}]])

    # Top
    fig.add_trace(go.Scatter(x= Bins, y=0*Bins, mode='lines+markers', name='Intervals',
                             line = dict(color='SkyBlue', width= 10),
                             marker=dict(size=15*np.ones(len(Bins)), opacity=1,
                                 color=np.random.choice(list(mcolors.CSS4_COLORS.keys()), size = len(Bins)))), row=1, col=1)
    fig.update_xaxes(showline=True, linewidth=1, linecolor='Lightgray', mirror=True,
                     showgrid=True, gridwidth=1, gridcolor='Lightgray',
                     tickvals=Bins, range=xLim, row=1, col=1)
    fig.update_yaxes(showline=True, linewidth=1, linecolor='Lightgray', mirror=True,
                     showticklabels=False, range= [-.2, .2], row=1, col=1)
    fig.update_traces(marker_line_color= 'Black', marker_line_width=1.5, opacity=1, row=1, col=1)
    fig.update_layout(plot_bgcolor= 'white', barmode='stack', width = W)

    # Bottom
    Bins = [int(x) for x in Bins]
    Table = pd.cut(Inp[Feat], bins = pd.IntervalIndex.from_tuples([(x, y) for x, y in zip(Bins[:-1],Bins[1:])]))
    Table = Table.value_counts().to_frame('Count').reset_index().rename(columns = {'index':Feat})  
    Table = Table.sort_values(by = [Feat])
    Table[Feat] = Table[Feat].astype(str)
    Table['Percentage'] = np.round(100*Table['Count']/Table['Count'].sum(), 2)
    T = Table.copy()
    T['Percentage'] = T['Percentage'].map(lambda x: '%.2f' % x)

    Temp = []
    for i in T.columns:
        Temp.append(T.loc[:,i].values)

    fig.add_trace(go.Table(header=dict(values = list(Table.columns), line_color='darkslategray',
                                       fill_color= TableColors[0], align=['center','center'],
                                       font= dict(color='white', size=12), height=25),
                           columnwidth = [0.4, 0.2, 0.2],
                           cells=dict(values=Temp, line_color='darkslategray', 
                                      fill=dict(color= [TableColors[1], TableColors[1]]),
                                      align=['center', 'center'], font_size=12, height=20)), row=2, col=1)

    fig.update_layout(title={'text': '<b>' + Feat + '<b>', 'x':0.5,
                                     'y':0.90, 'xanchor': 'center', 'yanchor': 'top'})
    fig.show()

Age

Feat = 'Age'
Bins = np.array([15, 30, 45, 65, 85], dtype = int)
Feat_Dict.update({Feat:Bins})
df[Feat] = FeatBins(Inp = Data[Feat], Bins = Bins, replace = True)
BinPlot(Feat, Bins, xLim = [0, 100])
del Feat, Bins

Checking Account

$$\mbox{Checking Account} = \begin{cases} 0, &\mbox{None},\\ 1, &\mbox{Little},\\ 2, &\mbox{Moderate},\\ 3, &\mbox{Rich}. \end{cases}$$

Feat = 'Checking Account'
Feat_Dict.update({Feat: {'None':0, 'Little': 1, 'Moderate':2, 'Rich':3}})
df[Feat] = df[Feat].replace(Feat_Dict[Feat]).astype(int)
del Feat

Credit Amount

Feat = 'Credit Amount'
Bins = np.array([0, 1e3, 5e3, 1e4, 2e4], dtype = int)
Feat_Dict.update({Feat:Bins})
# df[Feat] = FeatBins(Inp = Data[Feat], Bins = Bins, replace = True)
# BinPlot(Feat, Bins, xLim = [-1e3, 21e3])
# del Feat, Bins

Duration

Feat = 'Duration'
Bins = np.array([0, 6, 12, 18, 24, 80], dtype = int)
Feat_Dict.update({Feat:Bins})
df[Feat] = FeatBins(Inp = Data[Feat], Bins = Bins, replace = True)
BinPlot(Feat, Bins, xLim = [-5, 85])
del Feat, Bins

Housing

$$\mbox{Housing} = \begin{cases} 0, &\mbox{Free},\\ 1, &\mbox{Rent},\\ 2, &\mbox{Own,} \end{cases}$$

Feat = 'Housing'
Feat_Dict.update({Feat: {'Free':0, 'Rent': 1, 'Own':2}})
df[Feat] = df[Feat].replace(Feat_Dict[Feat]).astype(int)
del Feat

Job

Feat_Dict.update({'Job':{0:'Unskilled and Non-Resident', 1:'Unskilled and Resident', 2:'Skilled', 3:'Highly Skilled'}})

Purpose

Temp = df.drop(columns = ['Purpose'])
df = pd.concat([Temp, pd.get_dummies(df['Purpose'])], axis = 1)
del Temp

Risk

$$\mbox{Risk} = \begin{cases} 0, &\mbox{Bad},\\ 1, &\mbox{Good}. \end{cases}$$

Feat = 'Risk'
Feat_Dict.update({Feat: {'Bad':0, 'Good':1}})
df[Feat] = df[Feat].replace(Feat_Dict[Feat]).astype(int)
del Feat

Saving Accounts

$$\mbox{Saving Accounts} = \begin{cases} 0, &\mbox{None},\\ 1, &\mbox{Little},\\ 2, &\mbox{Moderate},\\ 3, &\mbox{Quite Rich},\\ 4, &\mbox{Rich}. \end{cases}$$

Feat = 'Saving Accounts'
Feat_Dict.update({Feat: {'None':0, 'Little': 1, 'Moderate':2, 'Quite Rich':3, 'Rich':4}})
df[Feat] = df[Feat].replace(Feat_Dict[Feat]).astype(int)
del Feat

Sex

$$\mbox{Sex} = \begin{cases} 0, &\mbox{Female},\\ 1, &\mbox{Male}. \end{cases}$$

Feat = 'Sex'
Feat_Dict.update({Feat: {'Female':0, 'Male':1}})
df[Feat] = df[Feat].replace(Feat_Dict[Feat]).astype(int)
del Feat

with open(Path.split(".")[0] + '_Feat_Dict.pkl', 'wb') as fp:
    pickle.dump(Feat_Dict, fp, protocol=pickle.HIGHEST_PROTOCOL)
    
df.to_csv (Path.split(".")[0]+'_Num.csv', index = None, header=True)

Features with high variance

Moreover, high variance for some features can hurt our modeling process. For this reason, we would like to standardize features by removing the mean and scaling to unit variance.

def Feature_Normalize(Inp, Target = Target, FS = (12, 8), annot_kws = 12):
    X = Inp.drop(columns = [Target])
    # scaling data
    scaler = preprocessing.StandardScaler()
    X_std = scaler.fit_transform(X)
    X_std = pd.DataFrame(data = X_std, columns = X.columns)

    fig, ax = plt.subplots(2, 1, figsize=FS)
    ax = ax.ravel()
    CP = [sns.color_palette("OrRd", 20), sns.color_palette("Greens", X.shape[1])]
    Names = ['Variance of the Features', 'Variance of the Features (Standardized)']
    Sets = [X, X_std]
    kws = dict(label='Feature\nVariance', aspect=10, shrink= .3)
    for i in range(len(ax)):
        Temp = Sets[i].var().sort_values(ascending = False).to_frame(name= 'Variance').round(2).T
        _ = sns.heatmap(Temp, ax=ax[i], annot=True, square=True,  cmap = CP[i],
                        linewidths = 0.8, vmin=0, vmax=Temp.max(axis =1)[0],  annot_kws={"size": annot_kws},
                        cbar_kws=kws)
        _ = ax[i].set_yticklabels('')
        _ = ax[i].set_title(Names[i], weight='bold', fontsize = 14)
        del Temp
    Out = Inp.copy()
    Out[X.columns] = X_std.copy()
    return Out

df = Feature_Normalize(df, FS = (14, 8), annot_kws = 12)
df.to_csv (Path.split(".")[0]+'_STD.csv', index = None, header=True)

Data Correlations

def Correlation_Plot (Inp, Fig_Size = 12, annot_kws = 11):
    Correlation_Matrix = Inp.corr().round(2)
    mask = np.zeros_like(Correlation_Matrix)
    mask[np.triu_indices_from(mask)] = True
    for i in range(len(mask)):
        mask[i,i]=0
    fig, ax = plt.subplots(figsize=(Fig_Size, Fig_Size))
    sns.heatmap(Correlation_Matrix, ax=ax, mask=mask, annot=True, square=True, 
                cmap =sns.color_palette("Greens", n_colors=10), linewidths = 0.2, vmin=0, vmax=1,
                cbar_kws={'label': 'Correlation', "aspect":30, "shrink": .4},  annot_kws={"size": annot_kws})
    return Correlation_Matrix

def Feature_Correlation(Inp, Target = Target, annot_kws = 11, FS = (17, 16), rep_long_labels = False):
    Fig, ax = plt.subplots(figsize = FS)
    Temp = Inp.corr().round(2)
    Temp = Temp.loc[(Temp.index == Target)].drop(columns = Target).T.sort_values(by = Target).T
    _ = sns.heatmap(Temp, ax=ax, annot=True, square=True,  cmap =sns.color_palette("YlGn", n_colors=Temp.shape[1]),
                    linewidths = 0.8, vmin=0, vmax=1,
                    annot_kws={"size": annot_kws},
                    cbar_kws={'label': 'Correlation with %s' % Target,
                              "aspect":40, "shrink": .4, "orientation": "horizontal"})
    if rep_long_labels:
        labels = [x.replace(' ','\n').replace('\nof\n',' of\n')
                  for x in [item.get_text() for item in ax.get_xticklabels()]]
        _ = ax.set_xticklabels(labels)
        _ = ax.set_yticklabels('')
        


_ = Correlation_Plot (df, 12, annot_kws = 11)
# Feature_Correlation(df, annot_kws = 11, FS = (12, 16), rep_long_labels = True)

---

References

1. Kaggle German Credit Risk Dataset

---